REPLICA SYMMETRY-BREAKING IN NEURAL NETWORKS WITH MODIFIED PSEUDO-INVERSE INTERACTIONS

Replica symmetry breaking is studied in fully connected neural networks with modified pseudo-inverse interactions. The interaction matrix has an intermediate form between the Hebb learning rule and the pseudo-inverse one. At low temperatures there is a region of parameters where the replica-symmetric solution is stable while its entropy is negative. It indicates the existence of an alternative solution in which the replica symmetry is broken. The one-step replica symmetry-breaking solution is found and its properties are analysed.